Restoration Ecology, TUM School of Life Sciences, Technical University of Munich, Germany

* Corresponding author: markus1.bauer@tum.de

Open Research: Data and code are permanently available on Zenodo under: https://doi.org/10.XXXX.

Max 8000 words

Abstract (max 300 words)

Keywords

dike grassland

embankment

species composition

environmental filter

1 Introduction

“Ecological theory can point restoration toward important processes that need manipulation […]. However, for this information to be relevant, restoration ecology needs to employ evidence-based assessments […]” (Suding 2011) –> No evidence for theory with our project

We asked following question to evaluate the restoration success after four years:

  1. How close is the vegetation to the reference state?

  2. How strong differs the Favourable Conservation Status (FCS) among the seed mixture-substrate combinations?

  3. How strong differs the persistence of the sown species among the seed mixture-substrate combinations?

2 Material & Methods

Experimental design

The seed mixture-substrate combination were tested with an experiment on a dike of the River Danube in SE Germany, which was established in March 2018 (Figure ??), 314 m above sea level (asl); WGS84 (lat/lon), 48.83895, 12.88412). The climate is temperate-suboceanic with a mean annual temperature of 8.4 °C and precipitation of 984 mm (Deutscher Wetterdienst 2021). During the study, three exceptionally dry years (2018–2020) occurred (Climate Data Center of the German Meteorological Service 2022a; Climate Data Center of the German Meteorological Service 2022b), and three minor floods which did not reach the plots Bayerisches Landesamt für Umwelt (2021b) (Table ??). Regional sand (0–4 mm) was used to lean the substrate and the soil was taken from the dike and a with an excavator the substrates were mixed and the plots prepared. The target vegetation were lowland hay meadows (Arrhenatherion elatioris, CM01A) and calcareous grassland (Cirsio-Brachypodion pinnati, DA01B code of the EuroVegChecklist, Mucina et al. (2016) ). The species pool of hay meadows and calcareous grasslands consisted of 55 and 58 species , respectively. The seeds were received from a commercial producer of autochtonous seeds. From the species pools, 20 species were randomly selected for each plot. The seed mixtures always contained seven grasses (60 wt% of total seed mixture), three legumes (5%) and ten herbs (35%) (Table ??). The community-weighted means (CWM) of functional traits differed between the seed mixture types (Table ??) . The management started with a cut without hay removal and a mowing height of 20 cm in August 2018, followed by ‘normal’ deep cuts with hay removal in July 2019 and 2020. In October 2018, Bromus hordeaceus was seeded as a nursery grass to provide safe sites under drought conditions.

We used 288 plots of the size 2.0 m × 2.0 m which were distributed over the north and south exposition of the dike and arranged in six blocks (= replicates). The experiment used a split plot design combined with randomized complete block design (Figure ??). The split plot was cuased by testing all 24 treatments on both sides of the dike. We tested three substrates with 0%, 25% and 50% sand admixture and two soil depths (15 vs. 30 cm). Below the substrate, a drainage layer of 5 cm consisting of gravel (0–16 mm) was installed. The sand admixture changed the soil texture, increased the C/N ratio, but reduced the ratio of Calciumcarbonat, and hardly changed the pH (Table ??). The two seed mixture types were tested with two seeding densities (4 vs. 8 g m^-2 ). The vegetation was surveyed in June or July 2018–2021 (Braun-Blanquet, 1928/1964). Establishment rates of species

Statistical analyses

We performed all analyses in R [Version 4.2.2; R Core Team (2022)], with the functions ‘blme’ (based on ‘lme4’) for BLMM Chung et al. (2013); and ‘DHARMa’ for model evaluation (Hartig 2022).

We calculated Bayesian linear mixed-effects models (BLMM) with the random effect ‘plot’ and used the restricted maximum-likelihood estimation (REML), the optimiser Nelder-Mead and, for the random effect, the Wishart prior. To identify the final model, we first reviewed the residual diagnostics of the candidate models and subsequently compared the remaining models using the Akaike information criterion adjusted for a small sample size (AICc) and chose the most parsimonious model (Appendix A4). Finally, we calculated the marginal and conditional coefficients of determination (R²m, R²c) and the 95% confidence intervals of the response variables.

3 Results

4 Discussion

Text

5 Conclusions

Text

Acknowledgements

We would like to thank our project partners Dr. Markus Fischer, Frank Schuster, and Christoph Schwahn (WIGES GmbH) and Klaus Rachl and Stefan Radlmair (Regierung von Niederbayern) for numerous discussions on restoration and management of dike grasslands. Field work was supported by Clemens Berger and Uwe Kleber-Lerchbaumer (Wasserwirtschaftsamt Deggendorf). We thank Holger Paetsch, Simon Reith, Anna Ritter, Jakob Strak, Leonardo H. Teixeira, and Linda Weggler for assisting with the field surveys or soil analyses in 2018–2020. The German Federal Environmental Foundation (DBU) supported MB with a doctoral scholarship.

Author contribution

JH and JK designed the experiment. JH did the surveys in the years 2018–2020 and MB in 2019 and 2021. MB did the analyses and wrote the manuscript. JK and JH critically revised the manuscript.

Open research

Data and code is available on Zenodo:

Model evaluation is stored on GitHub: github.com/markus1bauer/2023_danube_dike_experiment

Funding

MB is funded by a doctoral scholarship of the German Federal Environmental Foundation (DBU) (No. 20021/698). The establishment of the experiment and the vegetation surveys were financed by the WIGES GmbH in the years 2018–2020.

References

Bates, D., Machler, M., Bolker, B., & Walker, S. 2015. Fitting linear mixed-effects models using lme4. 67:
Bayerisches Landesamt für Umwelt. 2021a. Discharge pfelling / donau.
Bayerisches Landesamt für Umwelt. 2021b. Discharge pfelling / donau.
Chung, Y., Rabe-Hesketh, S., Dorie, V., Gelman, A., & Liu, J. 2013. A Nondegenerate Penalized Likelihood Estimator for Variance Parameters in Multilevel Models. Psychometrika 78: 685–709.
Climate Data Center of the German Meteorological Service. 2022a. Annual station observations of air temperature at 2 m above ground in °c for germany. Version v21.3: Station metten.
Climate Data Center of the German Meteorological Service. 2022b. Annual station observations of precipitation in mm for germany. Version v21.3: Station metten.
Deutscher Wetterdienst. 2021. Langjähriges mittel der wetterstation metten 1981-2010.
Hartig, F. 2022. DHARMa: Residual diagnostics for hierarchical (multi-level / mixed) regression models.
Mucina, L., Bültmann, H., Dierßen, K., Theurillat, J.-P., Raus, T., Čarni, A., Šumberová, K., Willner, W., Dengler, J., García, R.G., Chytrý, M., Hájek, M., Di Pietro, R., Iakushenko, D., Pallas, J., Daniëls, F.J.A., Bergmeier, E., Santos Guerra, A., Ermakov, N., Valachovič, M., Schaminée, J.H.J., Lysenko, T., Didukh, Y.P., Pignatti, S., Rodwell, J.S., Capelo, J., Weber, H.E., Solomeshch, A., Dimopoulos, P., Aguiar, C., Hennekens, S.M., & Tichý, L. 2016. Vegetation of Europe: hierarchical floristic classification system of vascular plant, bryophyte, lichen, and algal communities (R. Peet, Ed.). Applied Vegetation Science 19: 3–264.
R Core Team. 2022. R: A language and environment for statistical computing.

Tables

Table 1

Mischprobe aus mehreren Plots. Measured in < 2mm fraction

Table 2

Each plot received a individual set of 20 species with some restrictions to the number of species per functional group.

Figures

Figure 1

figure-1

Figure 1: figure-1

Figure 2

Figure 3

Figure 4

Figure 5

Fig. 5:

Figure 6

Fig. 6:

Supplementary Material

Table S1

Three dry years (2018–2020) and three minor floods (2018, 2019, 2021) occurred during the study period. Annual temperature, precipitation and discharge of River Danube near the study sites (2017–2021), based on weather station Metten [mean, 1981–2010; ID, 3271; WGS84, lat/lon, 48.85476, 12.918911; Climate Data Center of the German Meteorological Service (2022b), Climate Data Center of the German Meteorological Service (2022a)] and stream gauge Pfelling [ID, 10078000; WGS84, lat/lon, 48.87975, 12.74716; Bayerisches Landesamt für Umwelt (2021b)]. HQ2 = Highest discharge with a probability of occurrence every second year. HSW = Highest water level for shipping.

Table S2

Establishment rate

table-s2

Figure 2: table-s2

Table S3

See repository outputs/tables/table_a3_seedmixes.csv

Session Info

## R version 4.2.2 (2022-10-31 ucrt)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 10 x64 (build 22621)
## 
## Matrix products: default
## 
## locale:
## [1] LC_COLLATE=German_Germany.utf8  LC_CTYPE=German_Germany.utf8   
## [3] LC_MONETARY=German_Germany.utf8 LC_NUMERIC=C                   
## [5] LC_TIME=German_Germany.utf8    
## 
## attached base packages:
## [1] stats     graphics  grDevices datasets  utils     methods   base     
## 
## other attached packages:
##  [1] flextable_0.8.3 forcats_0.5.2   stringr_1.5.0   dplyr_1.0.10   
##  [5] purrr_0.3.5     readr_2.1.3     tidyr_1.2.1     tibble_3.1.8   
##  [9] ggplot2_3.4.0   tidyverse_1.3.2 knitr_1.41      bookdown_0.31  
## 
## loaded via a namespace (and not attached):
##  [1] httr_1.4.4          sass_0.4.4          bit64_4.0.5        
##  [4] vroom_1.6.0         jsonlite_1.8.4      here_1.0.1         
##  [7] modelr_0.1.10       bslib_0.4.1         assertthat_0.2.1   
## [10] highr_0.9           renv_0.16.0         googlesheets4_1.0.1
## [13] cellranger_1.1.0    yaml_2.3.6          gdtools_0.2.4      
## [16] pillar_1.8.1        backports_1.4.1     glue_1.6.2         
## [19] uuid_1.1-0          digest_0.6.30       rvest_1.0.3        
## [22] colorspace_2.0-3    htmltools_0.5.3     pkgconfig_2.0.3    
## [25] broom_1.0.1         haven_2.5.1         scales_1.2.1       
## [28] officer_0.4.4       tzdb_0.3.0          timechange_0.1.1   
## [31] googledrive_2.0.0   generics_0.1.3      ellipsis_0.3.2     
## [34] cachem_1.0.6        withr_2.5.0         cli_3.4.1          
## [37] magrittr_2.0.3      crayon_1.5.2        readxl_1.4.1       
## [40] evaluate_0.19       fs_1.5.2            fansi_1.0.3        
## [43] xml2_1.3.3          tools_4.2.2         data.table_1.14.6  
## [46] hms_1.1.2           gargle_1.2.1        lifecycle_1.0.3    
## [49] munsell_0.5.0       reprex_2.0.2        zip_2.2.2          
## [52] compiler_4.2.2      jquerylib_0.1.4     systemfonts_1.0.4  
## [55] rlang_1.0.6         grid_4.2.2          rstudioapi_0.14    
## [58] base64enc_0.1-3     rmarkdown_2.18      gtable_0.3.1       
## [61] DBI_1.1.3           R6_2.5.1            lubridate_1.9.0    
## [64] bit_4.0.5           fastmap_1.1.0       utf8_1.2.2         
## [67] rprojroot_2.0.3     stringi_1.7.8       parallel_4.2.2     
## [70] Rcpp_1.0.9          vctrs_0.5.1         dbplyr_2.2.1       
## [73] tidyselect_1.2.0    xfun_0.35